Absird Frequently Asked Questions


How does the program work?

    The program works by reading in a depth image.  It then
    tiles a pattern by varying amounts according to the depth at
    each pixel.  Where the image is close to the viewer, the
    pattern repeats rapidly.  Where the image is far from the
    viewer, the tile repeats slowly.


What's a "depth" image?  How can I make one or get one?

    A depth image is an image where the brightness of each pixel
    represents how far the image is from the viewer.  Most
    renderers can create them.  I include some in my package.
    Without a renderer, it is nearly impossible to make them.
    In this program, the brighter the pixel, the closer it is to
    the viewer.


On what machines does the program run?

    I only have compiled in on an SGI running UNIX.  If you port
    it to another (non-Microsoft) platform, I would like to know
    about it.  It should not be hard to port.


Does it have to use that complicated formula? [depthToOffset()]

    Strictly speaking, no.  Any interpolation which changes the
    repeat pattern in accordance with the depth will give the
    "stereogram look".  But using the real formula allows actual
    distance values to be applied and seen.  Flat areas from the
    depth image appear flat in the autostereogram.


Does it really have to anti-alias?

    Yes, it does.  It's amazing, but your eyes are quite
    sensitive to this effect.  Let's take a simple case.  Let's
    suppose that you've got a stereogram which you view at a
    distance of 15 inches, and you want the image to appear from
    15 to 25 inches into the screen.  Also suppose you're looking
    at a screen which has 72 pixels per inch.  That stereogram
    will have a repeating pattern which varies from 90 pixels to
    112.5 pixels in width.  That's a mere 22 depth values
    represented.  You can see these jumps.  If you anti-alias,
    it's smoother.


What's this "zeroDepth" nonsense?

    A common theme in stereograms is to have an object floating
    in front of a faraway wall.  So I put in this feature.  The
    zeroDepth value will be the depth on pixels which are zero.
    If you make it different than the minDepth, then the object
    will float in front of a wall at the zeroDepth distance.


Can I make "cross-eyed" stereograms by just negating the values?

    No, it's not that simple.  If I get better at seeing
    "cross-eyed" stereograms, and there is sufficient interest,
    I'll add an option for cross-eyed mode.


Why is the output stereogram *larger* than the input?

    Each of your eyes sees a copy of the stereogram.  The depth
    appears only in the overlap of these two copies.  So the
    depth is smaller than the stereogram by one tile.


Where to I get the jpeg image libraries & include files?

    The library and all the source for it are available at
    ftp.uu.net (Internet address 192.48.96.9) in the directory
    graphics/jpeg.  Very nice package.  (Hey, IJG, when does it
    get ported to C++?)


How can I use a different image library package?

    You'll need to write 6 routines:

        openJPEGRead()
        openJPEGWrite()
        readJPEGScanline()
        writeJPEGScanline()
        closeJPEGRead()
        closeJPEGWrite()

    which mimic the ones I've written but read & write your
    preferred image file type.  In theory, you shouldn't need to
    do anything else, but I haven't actually tried this, so maybe
    I did something wrong.  I tried to encapsulate it all....


How do I get in touch with you an ask you more questions?

    Keith Goldfarb              Rhythm & Hues               
    keith@rhythm.com            http://www.rhythm.com/~keith
    I got nothing.  Too bad.                                Some pink toes
    But I'm happy 'cause that's all I have.                 Some black toes